Forecasts to your front door

Hyperlocal predictions are the future of weather forecasting

A quick check of your phone's weather app shows that it's 70 degrees and sunny. But when you step outside, it's drizzling and there's a slight chill in the air.

Weather forecasts are better than they've ever been, but those disparities can still happen in a world where meteorologists make their calls for areas as wide as 20 miles across.

The future may be different. Hyperlocal forecasts, detailed down to areas as narrow as your neighborhood corner, or even your front door, might make significant advances with the help of artificial intelligence and machine learning.

“People don’t go on a run through their town; they run through their neighborhood,” said Matt Taylor, a meteorologist with Raytheon Intelligence & Space, a Raytheon Technologies business. “A weather forecast is kind of like a handshake, so if your app tells you it’s snowing outside, but it’s not, then you lack trust in it and you don’t feel inclined to use it and you don’t believe it.”

The AMS celebrated its 100th anniversary in 2020 with the theme, “The AMS Past, Present and Future: Linking Information to Knowledge to Society.” Raytheon Technologies was there to demonstrate some of its current and future weather technologies, some of which may help realize hyperlocal forecasting. The company is investing in research on how to use machine learning, analytics and artificial intelligence to improve the accuracy of weather forecasting.

One Raytheon Intelligence & Space project is trying to use machine learning to shorten the time it takes to predict tornadoes. Using massive amounts of data, including radar data, geography and other factors, researchers believe they could help meteorologists warn the public of tornado formations much more quickly.

“Every additional minute of lead time increases opportunities to take shelter and protect property,” said Nick Powell, a Raytheon Intelligence & Space meteorologist and engineer.

Raytheon Intelligence & Space is working with a company to investigate crowdsourcing data from smartphones, cell towers, cars, buildings and Internet-of-Things devices to create a very accurate, hyperlocal forecast.

A small radar called Skyler could help bring about hyperlocal forecasting. Instead of one, massive radar installation, it consists of smaller, one-meter square Active Electronically Scanned Array, or AESA, software-defined radar units.

It gives “you highly localized information like ground fog or flash-flooding in remote or urban locations; things that today’s radars can’t do,” said Michael Dubois, Raytheon Intelligence & Space Skyler lead.

Another internal research and development program the company is investigating correlates weather warnings by the National Weather Service and broadcasters and the resulting public actions. It's using data from social media posts and traffic cameras in the affected areas to see if and when people evacuate, take shelter or ignore the warnings.

Raytheon Intelligence & Space already provides advanced forecasting technologies, including the Advanced Weather Interactive Processing System, or AWIPS. It helps NOAA's National Weather Service issue forecasts, watches and warnings hours or even days before storms cross the horizon.

Forecasters at more than 100 NWS offices and centers nationwide use AWIPS hardware and software to monitor, organize, visualize and distribute weather data from thousands of sensors and sources, view high-detail charts, graphs and maps and issue weather forecasts, watches and warnings.

“Meteorologists have a myriad of data sources to consider when making forecasts,” said Shawn Miller, Raytheon Intelligence & Space civil space and weather technical director. “Wind speed, air pressure and temperature are familiar to most people, but detailed radar data, infrared satellite images and hydrological river data can be as important.”

The company also operates and maintains the Joint Polar Satellite System Common Ground System, a global system of ground antennas and high-performance computers that process more than 11 terabytes of weather information every day.

The National Weather Service reports that 85 percent of the data flowing into American weather forecast models comes from “polar orbiters,” providing the public with improved, long-range forecasting up to seven days out.

“Our forecasting ability has dramatically improved over the last decade and polar satellites deserve much of the credit,” said Anthony Serhal, Raytheon Intelligence & Space JPSS CGS program manager. "The system supports the scientific community and citizens alike."

The company has a history in the field, working on the first weather radars.

“When it comes to weather radar, we’ve been there since the very beginning – since the get-go,” Powell said.

In 1957, Raytheon Technologies produced the Weather Bureau’s WSR-57 radar, short for Weather Surveillance Radar. It was the first generation of weather radars manufactured solely for a national warning network.

“It was dependable and reliable, and I was still using one in 1973 when I was in the Air Force,” Powell said. “We’re actually displaying a WSR-57 that’s on loan from the Limon Heritage Museum at the 2020 AMS meeting…I wouldn’t be surprised if a tear or two is shed when people see it.”